researcharticle literaturereviewandcomparisonoftwo ...botox1(allergan, inc. irvine,...

RESEARCH ARTICLE

Literature Review and Comparison of TwoStatistical Methods to Evaluate the Effect ofBotulinum Toxin Treatment on Gait inChildren with Cerebral PalsyAngela Nieuwenhuys1,2*, Eirini Papageorgiou1,2, Todd Pataky3, Tinne De Laet4,Guy Molenaers5,6, Kaat Desloovere1,2

1 Department of Rehabilitation Sciences, KU Leuven, Leuven, Belgium, 2 Clinical Motion AnalysisLaboratory, University Hospitals Leuven, Leuven, Belgium, 3 Department of Bioengineering, ShinshuUniversity, Ueda, Japan, 4 Faculty of Engineering Science, KU Leuven, Leuven, Belgium, 5 Department ofDevelopment and Regeneration, KU Leuven, Leuven, Belgium, 6 Department of Orthopedics, UniversityHospitals Leuven, Leuven, Belgium

* [email protected]

Abstract

Aim

This study aimed at comparing two statistical approaches to analyze the effect of Botulinum

Toxin A (BTX-A) treatment on gait in children with a diagnosis of spastic cerebral palsy

(CP), based on three-dimensional gait analysis (3DGA) data. Through a literature review,

the available expert knowledge on gait changes after BTX-A treatment in children with CP is

summarized.

Methods

Part 1—Intervention studies on BTX-A treatment in children with CP between 4–18 years

that used 3DGA data as an outcome measure and were written in English, were identified

through a broad systematic literature search. Reported kinematic and kinetic gait features

were extracted from the identified studies. Part 2—A retrospective sample of 53 children

with CP (6.1 ± 2.3years, GMFCS I-III) received 3DGA before and after multilevel BTX-A

injections. The effect of BTX-A on gait was interpreted by comparing the results of paired

samples t-tests on the kinematic gait features that were identified from literature to the

results of statistical parametric mapping analysis on the kinematic waveforms of the lower

limb joints.

Results

Part 1–53 kinematic and 33 kinetic features were described in literature. Overall, there is no

consensus on which features should be evaluated after BTX-A treatment as 49 features

were reported only once or twice. Part 2—Post-BTX-A, both statistical approaches found

increased ankle dorsiflexion throughout the gait cycle. Statistical parametric mapping

PLOSONE | DOI:10.1371/journal.pone.0152697 March 31, 2016 1 / 17

OPEN ACCESS

Citation: Nieuwenhuys A, Papageorgiou E, Pataky T,De Laet T, Molenaers G, Desloovere K (2016)Literature Review and Comparison of Two StatisticalMethods to Evaluate the Effect of Botulinum ToxinTreatment on Gait in Children with Cerebral Palsy.PLoS ONE 11(3): e0152697. doi:10.1371/journal.pone.0152697

Editor: Michel R. Popoff, Institute Pasteur, FRANCE

Received: January 8, 2016

Accepted: March 17, 2016

Published: March 31, 2016

Copyright: © 2016 Nieuwenhuys et al. This is anopen access article distributed under the terms of theCreative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in anymedium, provided the original author and source arecredited.

Data Availability Statement: All relevant data areavailable via Figshare (https://dx.doi.org/10.6084/m9.figshare.2055729.v1).

Funding: AN is supported by an 'Onderzoekstoelage'(OT) of KU Leuven university (OT/12/100). EP issupported by the MD Paedigree project, a Model-Driven Paediatric European Digital Repository,partially funded by the European Commission underFP7 - ICT Programme (grant agreement no: 600932,http://www.md-paedigree.eu). The funders had norole in study design, data collection and analysis,decision to publish, or preparation of the manuscript.

http://crossmark.crossref.org/dialog/?doi=10.1371/journal.pone.0152697&domain=pdf

http://creativecommons.org/licenses/by/4.0/

https://dx.doi.org/10.6084/m9.figshare.2055729.v1

https://dx.doi.org/10.6084/m9.figshare.2055729.v1

http://www.md-paedigree.eu

analyses additionally found increased knee extension during terminal stance. In turn, fea-

ture analyses found increased outtoeing during stance after BTX-A.

Conclusion

This study confirms that BTX-A injections are a valuable treatment option to improve gait

function in children with CP. However, different statistical approaches may lead to different

interpretations of treatment outcome. We suggest that a clear, definite hypothesis should be

stated a priori and a commensurate statistical approach should accompany this hypothesis.

IntroductionPathological gait is one of the most striking characteristics in children with cerebral palsy (CP)[1]. When spasticity, weakness or other CP-related motor impairments manifest during walk-ing, they may significantly restrict patients at the level of ‘activities’ and ‘participation’ of theInternational Classification of Functioning, Disability and Health [2]. There is a wide variety oftreatments which can be applied to improve gait, ranging from conservative treatments such asphysiotherapy, orthotics, and Botulinum toxin A (BTX-A) injections to surgical interventionssuch as single event multilevel surgery and selective dorsal rhizotomy. Depending on a patient’ssymptoms and age, different treatment modalities might be appropriate.

Gait changes after treatment are often objectively quantified by using three-dimensionalgait analysis (3DGA). 3DGA provides a large amount of multivariate kinematic and kineticwaveforms, which have proven to be highly valuable during the clinical decision-making pro-cess[3], [4]. Researchers and clinicians are challenged to extract and analyze the clinically rele-vant information from this large amount of data.

In general, specific, directed hypotheses are not stated prior to data collection. For example,a typical null hypothesis may state that there are no differences between the knee and anklekinematics of children with CP pre- and post-treatment. As a consequence, the full gait cyclesof the knee and ankle joints should ideally be considered in statistical analysis because thehypothesis implicitly pertains to the full gait cycle and to all knee/ankle kinematic variables.

In the literature however, several intervention studies that have examined the effect of treat-ment on gait, have reduced the amount of 3DGA data a priori by analyzing a number of spe-cific kinematic or kinetic gait features, which refer to a specific time instant of a gait cycle, e.g.peak values [5–9]. The rationale behind the selection of these features is often unclear becausethose specific variables rarely appear explicitly in hypotheses. Most likely, they were chosenbased on available clinical expert knowledge, through literature search, or potentially afterviewing the data. Depending on the available clinical expertise, reducing this large amount ofdata a priori or post-hoc may introduce bias and potential clinically relevant information couldbe overlooked. Furthermore, by conducting statistical testing on multiple dependent gait fea-tures in relatively small sample sizes, the risk of detecting a false positive outcome (type I error)increases. In turn, a Bonferroni correction, which is often applied to deal with this risk, willincrease the probability of obtaining a false negative result (type II error)[10].

In the past years, a promising approach called statistical parametric mapping (SPM) wasintroduced to the field of biomechanics [11], [12]. SPM is a statistical method able to performhypothesis testing on kinematic and kinetic data in a continuous manner, thereby making apriori data reduction for non-directed hypotheses redundant. Additionally, it also takes intoaccount the dependency between different time instances of the gait cycle [11], [12]. SPM hasalready been used for example to evaluate whether electromyography (EMG) time-series of

Two Statistical Methods for Gait Analysis

PLOS ONE | DOI:10.1371/journal.pone.0152697 March 31, 2016 2 / 17

Competing Interests: The authors have declaredthat no competing interests exist.

four lower limb muscles during the stance phase are different between children and adults [13].So far, it has not yet been used to evaluate the outcome of treatment in CP.

After a thorough search of the available literature, following the inclusion criteria describedin the methods section of this paper, 223 peer reviewed scientific papers that evaluated theeffect of treatment on gait in children with CP based on the analysis of kinematic or kinetic gaitfeatures, were identified. Some of the gait features that were analyzed in these papers to quan-tify the effect of treatment on gait were recurrent in many studies, while some others were onlyreported a small number of times. Furthermore, it appeared that the definitions of several fea-tures were somewhat unclear, making it difficult for researchers to reproduce or confirm theresults. An example is the feature ‘hip extension during terminal stance’. It is not clear whetherthis feature refers to a specific time instance of the gait cycle, or whether it refers to the meanvalue or peak value of the hip during a phase of gait. Furthermore, the phase ‘terminal stance’could potentially be defined differently across various studies because patients presenting witha pathological gait might not have a typical 60/40 ratio for stance and swing phase or might notdisplay a typical heel rise in case they do not reach a flat foot position[14], [15].

To assess the value of SPM analysis as an alternative approach to feature analysis withregard to the interpretation of treatment outcome based on 3DGA in children with CP, thisstudy will focus on one treatment modality, namely BTX-A injections. BTX-A is used to treatspasticity and has been proven to improve function and delay deterioration towards fixed mus-cle contractures or bony deformities[16], [17]. After tone reduction, 3DGA can highlight achild’s ability to alter their gait pattern. It can also detect to what extent other clinical motorsymptoms such as postural instability or muscle weakness may contribute to the pathologicalgait pattern [18].

The aim of this study was twofold. First, the available expert knowledge on gait patternchanges in children with CP after BTX-A treatment was summarized. By performing a system-atic literature search, an overview of gait features that have been frequently reported in litera-ture and that have been shown to be responsive to BTX-A treatment was created. Secondly, theeffect of multi-level BTX-A treatment on gait in a retrospective sample of children with CP wasevaluated by comparing the results of the frequently reported feature analysis to the results ofSPM analyses on the kinematic waveforms of the lower limb joints. It was expected that SPMwould be judged as a valuable alternative statistical approach to describe the effect of BTX-Aon gait in a wider and more unbiased perspective than the traditional feature analysis.

Material and MethodsEthical approval for this project was granted by the Medical Ethical Committee of UniversityHospitals Leuven, reference s56036. All patient information was anonymized prior to statisticalanalysis. Two major methodological parts were related to the main study goals. The first partinvolved a literature search to define and select reported gait features that quantify the effect ofBTX-A treatment. The second part was an experimental outcome study, comparing SPM anal-ysis to feature analysis in order to interpret the outcome of 3DGA pre- and post-BTX-A treat-ment in a group of children with CP.

Literature searchA broad systematic search in the databases of Pubmed, Embase, Cinahl, and Web Of Sciencewas performed. Key words included cerebral palsy, diplegia, hemiplegia, quadriplegia, gaitanalysis, locomotion, walking, gait, feature, parameter, variable, and characteristic. Relevantwildcard symbols were used to ensure all key word variations were searched and if possible,searches were limited to human studies, age, and language. After removal of duplicates,



references were screened based on title and abstract. Eligibility of full-text papers was thenassessed based on the following inclusion criteria: (a) intervention studies evaluating the effectof treatment on gait, using instrumented 3DGA; (b) the experimental population consisting forat least 80% out of children between the ages of 4–18 years with a diagnosis of the spastic typeof CP; (c) a definition of kinematic and/or kinetic features, including at least joint angles ormoments; (d) English full text availability in a Belgian library or at request to the author. Weexcluded case series, literature reviews, and intervention studies that only reported indices(such as GPS, GGI, etc.), EMG features, spatio-temporal parameters or children with dystonia.

The literature search explored all papers that have reported any type of treatment modalityto improve gait in children with CP, using 3DGA as an outcome measurement tool. For thepurpose of this study, only intervention studies evaluating BTX-A treatment were selected. Allkinematic and kinetic features of the papers identified during the review, were extracted. Subse-quently, features with different terminology were grouped together in case they had a similarmeaning (e.g. minimal hip angle in sagittal plane during stance and maximum hip extensionduring the gait cycle). Apart from the initial literature search, two reviewers completed eachstep of the review process and a third reviewer was consulted in case of disagreement. The firstsearch was conducted on December 9, 2013 and it was updated on October 28, 2015.

Experimental outcome studyPatients and treatment characteristics. Patients were retrospectively selected from the

database of the Clinical Motion Analysis Laboratory of University Hospital Pellenberg, Leuven,Belgium. We considered children with CP who attended the hospital for BTX-A treatmentbetween 2004 and 2014. Children eligible for this study met following inclusion criteria: (a) agebetween 4–18 years, (b) a predominantly spastic diagnosis of CP, (c) walking with or withoutassistance of aids (GMFCS I-III), (d) BTX-A injections had occurred in at least hamstrings andgastrocnemius muscles (e) a maximum of three months between the date of BTX-A injectionand the pre-3DGA, (f) at least 1 month and maximally 4 months between the date of BTX-Ainjection and the post-3DGA. Exclusion criteria were: symptoms of dystonia or ataxia and pre-vious orthopedic surgery.

In case multiple gait analysis sessions were available for one patient, a preference was givento the gait analyses that were collected before and after the first or second BTX-A treatmentbecause patients are likely to improve more after the first treatments [19]. All BTX-A injectionswere performed by a pediatric orthopedic surgeon and were part of an integrated multileveltreatment approach which was previously described by Molenaers et al[20]. BTX-A injectionswere administered under general (mask) anesthesia, applying a dilution of 100 units (U) ofBotox1 (Allergan, Inc. Irvine, USA) in 5ml of saline. The muscles selected for treatment wereinjected at multiple sites with a maximum of 50U per site and a minimal distance of 5 cmbetween injection sites. In accordance with the integrated approach, serial stretching casts, anincreased number of physiotherapy sessions, and increased use of orthoses were planned afterBTX-A treatment.

Data collection procedure. All gait analyses were conducted in the clinical motion analy-sis laboratory of the University Hospital Pellenberg, using a Vicon system (Oxford Metrics,Oxford, UK) and two AMTI force plates (Advanced Mechanical Technology Inc., Watertown,MA, USA). Children were asked to walk barefoot on a 10m walkway at a self-selected, comfort-able speed. Ten to fifteen infra-red VICON cameras captured the position of retroflectivemarkers, which were placed on the bony landmarks of the child according to the Plug-In-Gaitmarker model of Vicon (Oxford Metrics, Oxford, UK). Nexus software was used to define gaitcycles and estimate joint angles over the three anatomical planes. For children with bilateral



CP, both legs were eligible for analysis and for children with unilateral CP, only the affected legwas included. Per included leg, we analyzed the average of two gait trials. After a thoroughquality check of all included gait trials, the kinematic parameters reported in literature as wellas waveform data were exported using custom-made Matlab software (Mathworks, Inc. version2014a).

Statistical analysis. Only the kinematic joint angle features that were identified during theliterature review were considered in the analysis, because good quality kinetic data were notavailable for all participants. In case the definition of a feature in literature was unclear and wewere unable to recalculate it for our own data, it was excluded from the analysis. For eachselected feature, a paired samples t-test compared the included group of CP patients pre-treat-ment versus post-treatment using SPSS Statistics for Windows, version 20.0 (Armonk, NY).The overall probability of making a type I error was maintained at α = 0.05, by adjusting p-val-ues according to the Holm procedure (a stepwise Bonferroni correction) [21–23].

SPM was performed on the time-normalized gait cycles of the lower limb joint angles acrossthe three anatomical planes, taking into consideration the dependency of all points of each gaitcycle. A conservative Bonferroni correction of α = 0.01 across the five joints (pelvis, hip, knee,ankle, and foot) maintained a family-wise Type I error rate of 5%.

Firstly, an SPM two-tailed paired t-test compared the mean joint angles of the knee andankle in the sagittal plane as well as the foot progression angle before and after treatment. Astatistical parametric map or SPM{t} was computed, representing the traditional univariate t-statistic being calculated at each point of the gait cycle; this approach is termed ‘mass univari-ate’. Afterwards, Random Field Theory was used to calculate the critical threshold t abovewhich only 1% (α = 0.01) of equally smooth random data samples’ SPM{t} waveforms wouldbe expected to cross[24]. Whenever the experimentally observed SPM{t} exceeded this criticalthreshold, we computed a supra-threshold cluster probability that indicated a significant statis-tical difference between the pre-treatment and post-treatment analyses in that part of the gaitcycle. A p-value for each supra-threshold cluster was calculated to specify the probability ofdiscovering a cluster with identical temporal breadth when equally smooth random data wouldbe analyzed [24].

Secondly, an SPM paired Hotellings T2 was computed to compare the mean joint angles ofthe pelvis and hip joint before and after BTX-A treatment. The SPM paired Hotellings T2 sta-tistic takes into account the time dependency of all points of the gait cycle, as well as the covari-ance of joint kinematics over the three anatomical planes [25]. Post-hoc t-tests were performedif the Hotellings T2 test presented a statistically significant outcome, i.e. when the criticalthreshold t (α = 0.01) was exceeded. These post-hoc t-tests constituted a separate analysis ofthe pelvis or hip kinematics in the three planes. A full description of this workflow is describedin Robinson et al. [13]. All analyses were performed in Python (Python 2.7.2; EnthoughtPython Distribution, Austin, TX), using open-source SPM1D code (v.0.3; www.spm1D.org).

Results

Literature searchThe literature search yielded a total of 2531 titles and abstracts, which was reduced to a selec-tion of 26 papers that evaluated BTX-A treatment using 3DGA in children with CP (Fig 1)[5],[6], [9], [18], [26–47]. Fifteen papers reported the effect of BTX-A treatment to the gastrocne-mius muscle, sometimes in combination with the soleus muscle [5], [6], [26–28], [30], [31],[35], [37–39], [41–43], [46]. Besides BTX-A injections to the gastrocnemius, nine papers alsoincluded the hamstrings in a multi-level treatment [9], [18], [29], [33], [34], [36], [40], [44],[47]. Two papers focused solely on BTX-A injections to the hamstrings [32], [45]. The papers



http://www.spm1D.org

reported a median of five features (range 2–49). After features with a similar meaning weregrouped together, 53 kinematic features (S1–S5 Tables) and 33 kinetic features (S6–S8 Tables)were identified. Eleven kinematic features, which were ambiguously defined, could not beincluded in the statistical analysis of the experimental outcome study (part 2). Figs 2–6 list all42 features that were selected for statistical analysis in the experimental outcome study andshow for each of those features the number of papers that have reported it to be responsive toBTX-A treatment in children with CP.

In general, results are mixed. Almost half of all kinematic features were reported only once(n = 12) or twice (n = 12). On the other hand, the maximal dorsiflexion angle during stancewas reported in 23 papers and 21 of them found an improved maximal dorsiflexion angle post-BTX-A[5], [6], [9], [18], [26], [27], [29–31], [34–44], [47]. There is also consensus that dorsi-flexion in the ankle during the swing phase improves after BTX-A treatment of the gastrocne-mius and/or soleus muscle. Seven papers reported an increased maximal dorsiflexion angleduring swing[26], [29], [36], [39], [41–43] and four papers reported an increased dorsiflexionangle at 50% of the gait cycle[9], [18], [30], [34]. At the level of the knee, seven out of eightpapers, which reported the maximal knee flexion angle during swing, agreed that it was notinfluenced by BTX-A treatment[5], [6], [9], [18], [31–33], [40]. Knee flexion angle at initial

Fig 1. Workflow literature review. This figure describes the workflow which was followed to identify the 26papers that were included in the literature review. Based on title and abstract, 2008 papers were excluded.After assessing the full-texts, 300 additional papers were excluded using the a priori defined inclusion criteria.In the end, we identified 223 papers that reported on the outcome of treatment in children with CP by meansof 3DGA evaluations. Of those 223 papers, 26 reported on the outcome of BTX-A treatment and wereincluded for this study.

doi:10.1371/journal.pone.0152697.g001



contact and maximal knee extension angle during stance are reported eight[5], [9], [18], [31–33], [40], [47] and twelve[5], [6], [18], [27], [31–33], [37], [40], [42], [44], [45] times respec-tively, yet results are contradicting. Only three out of eight papers reported a decreased kneeflexion angle at initial contact[5], [32], [47] and five out of twelve papers reported an improvedknee extension angle during stance[18], [27], [32], [40], [45]. In the hip joint, six papersreported no significant effect of BTX-A injections to the hamstrings and/or gastrocnemiusmuscles on the maximal hip flexion angle during swing[9], [18], [32], [33], [40], [45].

Of the 33 kinetic features that we identified, 25 were reported only once (n = 17) or twice(n = 8). The features reported most often are the second peak in the ankle moment curve (dur-ing the second half of stance phase)[9], [18], [28], [34], [47] and the peak ankle power genera-tion during the gait cycle (maximum positive ankle power)[5], [6], [26], [34], [39], [44], [47].Three out of five studies agreed that the second peak internal plantarflexion moment increased

Fig 2. Pelvis across anatomical planes: Mean (°) and (SD(°)) of kinematic gait features pre- and post-BTX-A treatment (N = 73) compared to SPManalysis (N = 73) and findings from literature review. Panel (A) shows the SPM {T2} statistic (α = 0.01) as a function of the gait cycle. The critical threshold(wide dashes) was not exceeded, indicating no significant improvement of BTX-A treatment on the pelvic joint kinematics across the three anatomical planes.Panel (B) shows the mean (°) and (SD) of features extracted from literature. * indicates a significant difference between pre- and post-BTX-A treatmentbased on Holm’s adjusted p-value (α < 0.05); Max. = maximum. Panel (C) indicates results from literature review. Res/Rep shows the number of papers thatreported the feature to be responsive to BTX-A / number of papers that reported the feature.


Fig 3. Hip across anatomical planes: Mean (°) and (SD(°)) of kinematic gait features pre- and post-BTX-A treatment (N = 73) compared to SPManalysis (N = 73) and findings from literature review. Panel (A) shows the SPM {T2} statistic (α = 0.01) as a function of the gait cycle. The critical threshold(wide dashes) was not exceeded, indicating no significant improvement of BTX-A treatment on the hip joint kinematics across the three anatomical planes.Panel (B) shows the mean (°) and (SD) of features extracted from literature. * indicates a significant difference between pre- and post-BTX-A treatmentbased on Holm’s adjusted p-value (α < 0.05); ST = stance; SW = swing; Max. = maximum. Panel (C) indicates results from literature review. Res/Rep showsthe number of papers that reported the feature to be responsive to BTX-A / number of papers that reported the feature.




post-BTX-A[9], [18], [47]. Six out of seven studies that reported the maximal ankle power gen-eration did not change significantly after BTX-A treatment[5], [6], [26], [34], [39], [47].

Experimental outcome studyA total of 53 patients were included in this study. Patient characteristics are described inTable 1. The majority of patients were diagnosed with bilateral CP (n = 36) and GMFCS level I(n = 25). There was a median of 26.5 days (range 1–91 days) on average between the pre-3DGA and the date of BTX-A treatment. The median time between the BTX-A treatment

Fig 4. Knee in sagittal plane: Mean (°) and (SD(°)) of kinematic gait features pre- and post-BTX-A treatment (N = 73) compared to SPM analysis(N = 73) and findings from literature review. Panel (A) shows two graphs. The top graph shows the mean kinematics of the knee in the sagittal plane of 73included legs pre-BTX-A treatment (light grey) versus post-BTX-A treatment (dark gray). The bottom graph represents the SPM {T} statistic (α = 0.01) as afunction of the gait cycle. The critical threshold t = 3.425 (wide dashes) was exceeded at 41–59% and at 86% of the gait cycle, indicating a significantimprovement of BTX-A treatment on the knee joint kinematics in the sagittal plane. Panel (B) shows the mean (°) and (SD) of features extracted fromliterature. * indicates a significant difference between pre- and post-BTX-A treatment based on Holm’s adjusted p-value (α < 0.05); ST = stance; SW = swing;Max. = maximum. Panel (C) indicates results from literature review. Res/Rep shows the number of papers that reported the feature to be responsive toBTX-A / number of papers that reported the feature.


Fig 5. Ankle in sagittal plane: Mean (°) and (SD(°)) of kinematic gait features pre- and post-BTX-A treatment (N = 73) compared to SPM analysis(N = 73) and findings from literature review. Panel (A) shows two graphs. The top graph shows the mean kinematics of the ankle in the sagittal plane of 73included legs pre-BTX-A treatment (light grey) versus post-BTX-A treatment (dark gray). The bottom graph represents the SPM {T} statistic (α = 0.01) as afunction of the gait cycle. The critical threshold t = 3.454 (wide dashes) was exceeded between 0–2% and 22–100% of the gait cycle, indicating a significantimprovement of BTX-A treatment on ankle dorsiflexion during the gait cycle. Panel (B) shows the mean (°) and (SD) of features extracted from literature. *indicates a significant difference between pre- and post-BTX-A treatment based on Holm’s adjusted p-value (α < 0.05); GC = gait cycle; ST = stance;SW = swing; Max. = maximum. Panel (C) indicates results from literature review. Res/Rep shows the number of papers that reported the feature to beresponsive to BTX-A / number of papers that reported the feature.




session and the post-BTX-A 3DGA was 58 days (range 45–114 days). Seventy-three legs wereincluded in statistical analysis.

The median dosage of BTX-A injected into multiple sites of the hamstrings was 4U/kg bodyweight with a range of 2 to 6 U/kg body weight. A median dosage of 4 U/kg body weight wasalso administered to the gastrocnemius muscle, spread over different sites, with a range of 2 to7.5 U/kg body weight. Often iliopsoas, adductors, rectus femoris, soleus, and tibialis posteriorwere also included in the multilevel BTX-A treatment, though less frequently (Table 2). Inaccordance to the integrated treatment approach, all children received serial stretching castsfor the lower and/or upper legs for a period between 1 to 4 weeks. Children with unilateral CPalso received casts for both legs with the aim of obtaining a more symmetric gait pattern.

Fig 6. Foot progression angle: Mean (°) and (SD(°)) of kinematic gait features pre- and post-BTX-A treatment (N = 73) compared to SPM analysis(N = 73) and findings from literature review. Panel (A) shows two graphs. The top graph shows the mean kinematics of the foot progression angle of 73included legs pre-BTX-A treatment (light grey) versus post-BTX-A treatment (dark gray). The bottom graph represents the SPM {T} statistic (α = 0.01) as afunction of the gait cycle. The critical threshold t = 3.341 (wide dashes) was not exceeded, indicating no significant effect of BTX-A treatment on the footprogression angle. Panel (B) shows the mean (°) and (SD) of features extracted from literature. * indicates a significant difference between pre- and post-BTX-A treatment based on Holm’s adjusted p-value (α < 0.05). Panel (C) indicates results from literature review. Res/Rep shows the number of papers thatreported the feature to be responsive to BTX-A / number of papers that reported the feature.


Table 1. Patient Characteristics (N = 53).

Gender

Female 18

Male 35

Diagnosis

Bilateral CP 36

Unilateral CP 17

Mean age (at time of pre-3DGA) (years, SD) 6.1 (2.3)

Mean weight (at time of pre-3DGA) (kg, SD) 20.1 (7.0)

Mean height (at time of pre-3DGA) (cm, SD) 114.0 (14.6)

GMFCS

Level I 25

Level II 17

Level III 11

Walking aids during 3DGA

None 43

Support of one hand 1

Kayewalker 9

doi:10.1371/journal.pone.0152697.t001



Figs 2–6 describe the results for all statistical analyses. For the pelvis and hip joint, neitherof the statistical analyses found significant changes. SPM did find significant improvementspost-BTX-A-treatment between 41–59% and at 86% of the gait cycle, while the ankle dorsiflex-ion improved post-BTX-A treatment between 0–2% and 22–100% of the gait cycle. Figs 2–6also show that out of 42 features, 11 ankle joint features in the sagittal plane and the mean footprogression angle during stance were found to be significantly improved after BTX-Atreatment.

DiscussionIn this study, we tested the hypothesis that lower limb joint kinematics of children with a spas-tic diagnosis of CP would improve toward a more typical gait pattern post-BTX-A. We com-pared two statistical approaches. On the one hand, we analyzed kinematic gait features thathave previously been reported in literature and on the other hand we conducted SPM analyses.In summary, both approaches detected gait changes, mainly at the ankle. Both feature andSPM analyses concluded that no changes in gait occurred at the level of the pelvis and hip.After treatment, based on SPM analysis, we noted a significantly improved knee extension dur-ing stance, an earlier peak knee flexion during swing, and an increased ankle dorsiflexionthroughout most of the gait cycle. Post-BTX-A, feature analysis also highlighted improved dor-siflexion of the ankle at different time points of the gait cycle, and additionally an increasedouttoeing during stance of the foot in the transverse plane, but no effect at the knee. From liter-ature, we learned that results of BTX-A treatment based on feature analyses are generallymixed. Furthermore, feature definitions were not always clear enough to allow us to recalculatethem on our own data. For each joint, a more detailed discussion of our results compared to lit-erature is presented below.

Pelvis and hipThe SPM and feature analyses of the experimental outcome study did not highlight significanteffects of BTX-A treatment on the pelvic and hip kinematics in the three anatomical planes. Inthe same line, few papers in literature report a significant change of pelvic and hip kinematicspost-BTX-A treatment. Three out of five studies evaluating ‘mean pelvic tilt in the sagittalplane’, reported a significantly higher anterior tilt after treatment[32], [33], [44]. Corry et al.reported an increased anterior tilt as a source of concern after hamstrings injection if the psoaswas left untreated[32]. In the present study, 71% of treated limbs also received psoas injections,hence we did not expect an increased pelvic tilt after BTX-A. In the hip, literature frequentlyreported on three kinematic features, namely angle at initial contact, maximal hip extension

Table 2. Muscles treated with BTX-A (N = 73 treated limbs).

Muscles Number of limbs injected Median dose U/kg body weight (range)

Iliopsoas * 52 2 (1–3)

Adductors 37 1.5 (1–3)

Rectus femoris 11 1.5 (0.75–2)

Hamstrings 100 4 (2–6)

Gastrocnemius * 73 4 (2–7.5)

Soleus 18 2 (1–3)

Tibialis posterior 5 2 (1.5–2)

* Median dose and range are based on 71 limbs, as dosages were unavailable for two limbs.

doi:10.1371/journal.pone.0152697.t002



during stance, and maximal hip flexion during swing. Galli et al. evaluated BTX-A injections tothe gastrocnemius muscle and reported a slight deterioration towards an increased hip flexionthroughout the gait cycle, which emphasizes the need for a multilevel treatment approach[5].Depending on whether the hip flexors are included in the multilevel BTX-A treatment, themaximal hip extension in stance may significantly improve or not. This was demonstrated byDesloovere et al.[33], Svehlik et al.[44] and also by Papadonikolakis et al. who has reported anincrease in maximal hip extension during stance, only in the group of patients who received amulti-level BTX-A treatment[40]. However, it must be noted that neither of these studiesattempted to maintain the type I error rate at 0.05 by accounting for the covariance of multipledependent gait features, so a false positive result cannot be excluded.

KneeCompared to the pre-BTX-A condition, SPM analysis of the experimental data noted a signifi-cantly improved knee extension and an increased slope towards flexion during terminal stanceand pre-swing (between 41% - 59% of the gait cycle). This result may be related to the ham-strings injections. Spasticity of the hamstrings could impede a sufficient amount of knee exten-sion during stance and at the end of swing phase. After BTX-A injection into the hamstrings, areduction in tone can be achieved; hence improved knee extension during stance can beexpected. Additionally, an improvement in knee extension could also be expected after BTX-Ainjections to the gastrocnemius. On the contrary, no significant differences were found basedon the paired samples t-tests on seven gait features of the knee. Is it possible that the two fea-tures related to terminal stance and pre-swing, namely ‘maximal knee extension angle duringstance’ and ‘knee flexion angle at toe-off’, are insufficient to characterize these changes in theknee motion during gait post-BTX-A. At the level of the knee, changes in gait might be bettercharacterized by considering the terminal stance and pre-swing phase as a whole. SPM analysesof gait phases are easily interpretable and avoid the need of defining additional features whilefurther contributing to the problem of ‘multiple t-testing’. Alternatively, if feature analysis ispreferred, we may also conclude that it would be interesting to define an additional feature, forinstance knee flexion velocity during pre-swing.

In literature, results are mixed as five out of twelve papers found a significant increase inmaximal knee extension during stance[18], [27], [32], [40], [45]. It should be noted that allpapers who did not report an improved knee extension during stance either focused theBTX-A treatments solely on the triceps surae[5], [6], [31], [37], [42] or the post-3DGA evalua-tion was, on average, longer than one year[33], [44].

SPM analysis also indicated a significant change in a short phase of the knee kinematics dur-ing mid-swing, possibly indicating a shift in timing of achieving maximal knee flexion duringswing. Unfortunately, the definition of ‘amount of delayed knee flexion in swing’, which wasreported twice in literature with mixed findings, could not be interpreted and reproduced inour analysis[9], [18]. The only feature during swing that we analyzed was ‘maximal knee flex-ion angle during swing’, which was found to be unchanged after treatment. In literature, thisfeature was significantly increased in only one out of eight papers[5]. Galli et al. did not controlfor an increased type I error risk in their analysis of seventeen kinematic and kinetic features[5]. Nevertheless, their study solely investigated treatment of the gastrocnemius muscle, whilethe peak knee flexion during swing may be predominantly related to rectus femoris spasticity.In the current study, the average peak knee flexion during swing was around 60°, which is wellwithin the one standard deviation around the mean of our typically developing children. Thismay explain the fact that the rectus femoris was only injected in 15% of treated limbs. Of theother seven studies that reported no changes in the peak knee flexion during swing, only



Papadonikolakis et al. analyzed short-term changes after BTX-A in a patient group of whomsome also received rectus femoris injections[40]. However, it is unclear for how many patientsthe rectus femoris was included.

AnkleSPM analysis of the experimental data highlighted a significant increase in dorsiflexion at ini-tial contact after BTX-A treatment. From mid-stance through the remaining part of the gaitcycle, the dorsiflexion angle was shown to be significantly increased as well. Features evaluatedusing paired samples t-tests resembled the results of the SPM analysis. Also in literature, manypapers have provided evidence to support these results[5], [6], [9], [18], [26], [27], [29–31],[34–44], [47]. These results were expected as many CP children suffer from gastrocnemiusspasticity potentially leading to equinus gait, which is characterized by an abnormal plantar-flexion angle throughout the gait cycle. By means of BTX-A injections, the resulting tonereduction facilitates the motion towards dorsiflexion in stance and to clear the foot duringswing phase. In the context of an integrated spasticity treatment, the serial stretching castswhich were applied during the first weeks after BTX-A might further support this effect. Thestudy of Bottos et al. is the only paper not reporting a short-term beneficial effect of BTX-A onthe ankle joint kinematics[28]. However these results are probably due to a limited sample size(BTX-A groep, n = 5 and BTX-A plus casting group, n = 5).

FootAt the level of the foot, in contrary to SPM which found no effects, the paired samples t-test onthe experimental data revealed a significant improvement of the ‘mean foot progression angleduring the stance phase’. Because the motion of one joint is also co-varying (dependent) withthe motion at other joints, we applied a Bonferroni correction for the five SPM analyses toobtain a more appropriate alpha level. However, as was reported previously in literature, a Bon-ferroni correction might be too strict and thus produce an increased chance of false negatives(type II error) [10]. With regard to the feature analysis, we applied a Holm’s correction or step-wise Bonferroni correction for all features to maintain an alpha level of 0.05, with a lowerchance of false negatives (higher power) [21], [22], which might explain the different outcomesbetween both analysis approaches. Compared to literature, one study by Desloovere et al. didnot find a significant effect on the foot progression angle[33]. However, the post-3DGA evalua-tion in this study was on average one year and ten months post-BTX-A injections, making itunlikely to detect a therapeutic effect from the treatment. Two other studies also reported anincreased outtoeing after BTX-A[9], [18]. They both performed Bonferroni corrections, mak-ing a false positive result unlikely. In these two studies, as well as in our experimental study, allchildren received injections to the gastrocnemius along with a period of stretching casts to thelower legs. Hence more gait improvements can be expected in the distal joints such as the ankleand foot.

LiteratureApart from different statistical approaches and whether or not the Type I error is controlled,another factor that could account for differences in results among the included papers could bethe small sample sizes. With twelve papers evaluating a patient group of less than 20 patients, itis possible that a bias was present in the selected experimental population[5], [6], [27], [28],[31], [32], [34], [37], [44–47]. The timing of the post-BTX-A 3DGA ranged from two weeks tomore than one year across the included papers. This may result in heterogeneous conclusions,given the limited time frame in which BTX-A injections are effective and the importance to



acknowledge the effect of other parameters, such as age, on gait when timing of follow-upincreases. Naturally, the different clinical characteristics of experimental patient groups as wellas the diversity regarding the different muscles which were treated and the dosage of BTX-A,may also contribute to different outcomes. However a detailed analysis of all these factorswould require a thorough evaluation of the methodological quality of the papers, which wasbeyond the scope of the current study.

LimitationsLimitations of the current study need to be addressed. While creating an overview of the fea-tures from literature, rather subjective judgments were made on whether or not particular fea-tures could be merged. To avoid bias as much as possible all features were first independentlyjudged by two reviewers, after which a third reviewer was consulted in cases of disagreement.This task was also complicated because a number of features were not clearly defined and thusnot easily recalculated with our own data. Consequently, we were unable to include all reportedfeatures in the outcome study. Another limitation was the study population, which wasrecruited from the retrospective database of the University Hospital Pellenberg, Belgium,where every child is treated according to the same rehabilitation guidelines. As a consequence,generalization of results is limited and could be improved through multi-center studies. It wasreported by Molenaers et al. that the first two BTX-A treatments are most effective in increas-ing function[19]. In our intervention study, thirteen patients who had received a third BTX-Atreatment and one who had received a fourth treatment were included, which could have led toa reduced effect of BTX-A compared to earlier studies. A sample selection, only containingchildren after their first or second BTX-A treatment was not always possible because childrenwere often too young for a 3DGA when they were receiving their first treatments.

ConclusionIn conclusion, both the outcome of the current study as well as literature reports conclude thatBTX-A injections are a valuable treatment option to improve gait in children with CP. Theeffects were mainly observed at the ankle joint and to a lesser extent at the knee. Differentresults can be explained by various factors which have been illustrated in this study. The keyissue which we discussed in this study was the statistical analysis used to analyze 3DGA data.In literature, the effect on specific gait features is traditionally reported. Considering that overhalf of all extracted features were only reported once or twice for a list of 26 studies, it can beconcluded that there is no consensus on which features should be evaluated to assess the effectof BTX-A on the gait pattern of children with CP. Additionally, the risk of obtaining false posi-tive results (Type I error) quickly increases when multiple dependent gait features are analyzedand attempts to control this risk using a Bonferroni correction were in turn decreasing power(increasing the chance of Type II error).

Our study compared this frequently reported feature analysis to SPM. Our findings suggestthat both statistical methods might be appropriate to analyze kinematic and kinetic data toexamine the effect of BTX-A on gait. However, we suggest that a clear, definite hypothesisshould be stated a priori and an adequate, statistical approach should be selected to accompanythis hypothesis. When an analysis of features is preferred following a specific hypothesis, wenote that alternatives to the Bonferroni correction are available to deal with the risk of makinga Type I error[10], [21], [22]. The currently presented literature review could be a guide for fea-ture selection. When reporting features, care should be taken for an unambiguous feature defi-nition that is clinically meaningful. For example ‘maximal dorsiflexion during the gait cycle’could, depending on the clinical presentation of the child, occur during loading response, at



the beginning of the third rocker, or during swing phase and might thus not be very meaningfulto a clinician.

If it is difficult to specifically hypothesize on the direction and magnitude of changes post-treatment, SPM analysis might be preferred. SPM analysis allows you to analyze kinematic andkinetic waveforms as a whole, or particular gait phases, e.g. the swing phase for the knee joint.It has the advantage of making a priori data reduction redundant and taking into account thecovariance of all points of the gait cycle. Nevertheless, at this time, it is not possible to robustlyaddress the co-variation of different joints (e.g. knee and ankle).

For this study, we chose to illustrate the value of SPM by analyzing the effect of BTX-Atreatment on gait because treatment protocols are well standardized and its effect has oftenbeen reported in literature. Furthermore, the effect of BTX-A is evaluated rather quickly aftertreatment, ensuring little impact of other treatments or other factors such as age that may influ-ence gait. However, it would also be interesting to analyze the effects of other treatments inchildren with CP, such as selective dorsal rhizotomy or orthopedic surgery using SPM. SPMmight even be more appropriate for pathologies where kinematic and kinetic gait deviationspost-treatment have not yet been routinely recognized by clinicians. Non-directed hypothesesevaluated with SPM analysis may help to reduce the wealth of 3DGA data and aid the construc-tion of more specific, directed hypotheses in the future.

Supporting InformationS1 PRISMA Checklist.(PDF)

S1 TREND Checklist.(PDF)

S1 Table. Kinematic pelvic features extracted from 26 papers, identified through systematicliterature search.(PDF)

S2 Table. Kinematic hip features extracted from 26 papers, identified through systematicliterature search.(PDF)

S3 Table. Kinematic knee features extracted from 26 papers, identified through systematicliterature search.(PDF)

S4 Table. Kinematic ankle features extracted from 26 papers, identified through systematicliterature search.(PDF)

S5 Table. Kinematic features of the foot in the transverse plane extracted from 26 papers,identified through systematic literature search.(PDF)

S6 Table. Kinetic hip features extracted from 26 papers, identified through systematic liter-ature search.(PDF)

S7 Table. Kinetic knee features extracted from 26 papers, identified through systematic lit-erature search.(PDF)



http://www.plosone.org/article/fetchSingleRepresentation.action?uri=info:doi/10.1371/journal.pone.0152697.s001









S8 Table. Kinetic ankle features extracted from 26 papers, identified through systematic lit-erature search.(PDF)

AcknowledgmentsWe thank MSc. Frieke Ostijn and MSc. Jasmine Persoon for all their help during the systematicsearch of the literature and for aiding us during the patient selection process.

Author ContributionsConceived and designed the experiments: AN TP TDL GM KD. Performed the experiments:AN EP GM. Analyzed the data: AN EP TP. Wrote the paper: AN EP TP TDL GM KD.

References1. Ostensjø S., Carlberg E. B., and Vøllestad N. K., “Motor impairments in young children with cerebral

palsy: relationship to gross motor function and everyday activities.,” Dev. Med. Child Neurol., vol. 46,no. 9, pp. 580–589, 2004. PMID: 15344517

2. Rosenbaum P. and Stewart D., “TheWorld Health Organization International Classification of Function-ing, Disability, and Health: A Model to Guide Clinical Thinking, Practice and Research in the Field ofCerebral Palsy,” Semin. Pediatr. Neurol., vol. 11, no. 1, pp. 5–10, 2004. PMID: 15132248

3. Wren T. A. L., Otsuka N. Y., Bowen R. E., Scaduto A. A., Chan L. S., Sheng M., et al., “Influence of gaitanalysis on decision-making for lower extremity orthopaedic surgery: Baseline data from a randomizedcontrolled trial.,”Gait Posture, vol. 34, no. 3, pp. 364–9, 2011. doi: 10.1016/j.gaitpost.2011.06.002PMID: 21723131

4. Cook R. E., Schneider I., Hazlewood M. E., Hillman S. J., and Robb J. E., “Gait Analysis Alters Deci-sion-Making in Cerebral Palsy.,” J. Pediatr. Orthop., pp. 292–295, 2003. PMID: 12724589

5. Galli M., Cimolin V., Valente E. M., Crivellini M., Ialongo T., and Albertini G., “Computerized gait analy-sis of botulinum toxin treatment in children with cerebral palsy.,” Disabil. Rehabil., vol. 29, no. 8, pp.659–64, Apr. 2007. PMID: 17453987

6. Zwick E.B., Svehlik M., Steinwender G., and Linhart W.E., “Short-term effects of botulinum toxin A andserial casting on triceps surae muscle length and equinus gait in children with cerebral palsy,” Ces. aSlov. Neurol. a Neurochir., vol. 72, no. 6, pp. 553–558, 2009.

7. Dreher T., Vegvári D., Wolf S. L., Klotz M., Müller S., Metaxiotis D., et al., “Long-term effects after con-version of biarticular to monoarticular muscles compared with musculotendinous lengthening in chil-dren with spastic diplegia.,”Gait Posture, vol. 37, no. 3, pp. 430–5, Mar. 2013. doi: 10.1016/j.gaitpost.2012.08.020 PMID: 23018029

8. Marconi V., Hachez H., Renders A., Docquier P.-L., and Detrembleur C., “Mechanical work and energyconsumption in children with cerebral palsy after single-event multilevel surgery.,”Gait Posture, vol.40, no. 4, pp. 633–639, 2014. doi: 10.1016/j.gaitpost.2014.07.014 PMID: 25107323

9. Molenaers G., Desloovere K., De Cat J., Jonkers I., De Borre L., Pauwels P., et al., “Single event multi-level botulinum toxin type A treatment and surgery: similarities and differences.,” Eur. J. Neurol., vol. 8Suppl 5, pp. 88–97, Nov. 2001. PMID: 11851737

10. Knudson D., “Significant and meaningful effects in sports biomechanics research.,” Sports Biomech.,vol. 8, no. 1, pp. 96–104, 2009. doi: 10.1080/14763140802629966 PMID: 19391497

11. Pataky T. C., “Generalized n-dimensional biomechanical field analysis using statistical parametric map-ping,” J. Biomech., vol. 43, no. 10, pp. 1976–1982, 2010. doi: 10.1016/j.jbiomech.2010.03.008 PMID:20434726

12. Pataky T. C., Robinson M. a., and Vanrenterghem J., “Vector field statistical analysis of kinematic andforce trajectories,” J. Biomech., vol. 46, no. 14, pp. 2394–2401, 2013. doi: 10.1016/j.jbiomech.2013.07.031 PMID: 23948374

13. Robinson M. a., Vanrenterghem J., and Pataky T. C., “Statistical Parametric Mapping (SPM) for alpha-based statistical analyses of multi-muscle EMG time-series,” J. Electromyogr. Kinesiol., vol. 25, no.1, pp. 14–19, 2015. doi: 10.1016/j.jelekin.2014.10.018 PMID: 25465983

14. J. Perry,Gait analysis. Thorofare, NJ, Slack, 1992.

15. R. Baker, “Limitations of the conventional subdivision of the gait cycle,” inMeasuring walking, a hand-book of clinical gait analysis, 213AD, pp. 216–217.




http://www.ncbi.nlm.nih.gov/pubmed/15344517


http://dx.doi.org/10.1016/j.gaitpost.2011.06.002










http://dx.doi.org/10.1080/14763140802629966


http://dx.doi.org/10.1016/j.jbiomech.2010.03.008





http://dx.doi.org/10.1016/j.jelekin.2014.10.018


16. Graham H. K., Aoki K. R., Autti-Rämö I., Boyd R. N., Delgado M. R., Gaebler-Spira D. J., et al., “Recom-mendations for the use of botulinum toxin type A in the management of cerebral palsy.,”Gait Posture,vol. 11, no. 1, pp. 67–79, 2000. PMID: 10664488

17. Lukban M. B., Rosales R. L., and Dressler D., “Effectiveness of botulinum toxin A for upper and lowerlimb spasticity in children with cerebral palsy: a summary of evidence,” J. Neural Transm., vol. 116, no.3, pp. 319–331, 2009. doi: 10.1007/s00702-008-0175-8 PMID: 19142573

18. Desloovere K., Molenaers G., Jonkers I., De Cat J., De Borre L., Nijs J.,et al., “A randomized study ofcombined botulinum toxin type A and casting in the ambulant child with cerebral palsy using objectiveoutcomemeasures.,” Eur. J. Neurol., vol. 8 Suppl 5, pp. 75–87, 2001. PMID: 11851736

19. Molenaers G., Fagard K., Van Campenhout A., and Desloovere K., “Botulinum toxin A treatment of thelower extremities in children with cerebral palsy,” J. Child. Orthop., vol. 7, no. 5, pp. 383–387, 2013.doi: 10.1007/s11832-013-0511-x PMID: 24432099

20. Molenaers G., Desloovere K., Eyssen M., De Cat J., Jonkers I., and De Cock P., “Botulinum toxin typeA treatment of cerebral palsy: an integrated approach,” Eur. J. Neurol., vol. 6, pp. S51–S57.

21. McLaughlin M. J. and Sainani K. L., “Bonferroni, holm, and hochberg corrections: Fun names, seriouschanges to P values,” PMR, vol. 6, no. 6, pp. 544–546, 2014. doi: 10.1016/j.pmrj.2014.04.006 PMID:24769263

22. Holm S., “A Simple Sequentially Rejective Multiple Test Procedure,” Scand. J. Stat., vol. 6, no. 2, pp.65–70, 1979.

23. G. J., “Holm-Bonferroni sequential correction: an excel calculator (1.2).” 2013.

24. Adler R. and Taylor J., Random Fields and Geometry, 1st ed. New York: Springer-Verlag New York,2007.

25. Pataky T. C., “One-dimensional statistical parametric mapping in Python,” Comput. Methods Biomech.Biomed. Engin., vol. 15, no. 3, pp. 295–301, 2012. doi: 10.1080/10255842.2010.527837 PMID:21756121

26. Ackman J. D., Russman B. S., Thomas S. S., Buckon C. E., Sussman M. D., Masso P., et al., “Compar-ing botulinum toxin A with casting for treatment of dynamic equinus in children with cerebral palsy.,”Dev. Med. Child Neurol., vol. 47, no. 9, pp. 620–7, Sep. 2005. PMID: 16138670

27. Bang M. S., Chung S. G., Kim S. B., and Kim S.-J., “Change of dynamic gastrocnemius and soleusmuscle length after block of spastic calf muscle in cerebral palsy.,” Am. J. Phys. Med. Rehabil., vol. 81,no. 10, pp. 760–4, Oct. 2002. PMID: 12362116

28. Bottos M., Benedetti M. G., Salucci P., Gasparroni V., and Giannini S., “Botulinum toxin with and with-out casting in ambulant children with spastic diplegia: A clinical and functional assessment,” Dev. Med.Child Neurol., vol. 45, no. 11, pp. 758–762, 2003. PMID: 14580131

29. Boyd R. N., Graham J. E. A., Nattrass G. R., and Graham H. K., “Medium-term response characterisa-tion and risk factor analysis of botulinum toxin type A in the management of spasticity in children withcerebral palsy,” Eur. J. Neurol., vol. 6, no. 4, pp. S37–S45, Nov. 1999.

30. Boyd R. N., Pliatsios V., Starr R., Wolfe R., and GrahamH. K., “Biomechanical transformation of thegastroc-soleus muscle with botulinum toxin A in children with cerebral palsy.,” Dev. Med. Child Neurol.,vol. 42, no. 1, pp. 32–41, 2000. PMID: 10665973

31. Corry I. S., Cosgrove A. P., Duffy C. M., McNeill S., Taylor T. C., and Graham H. K., “Botulinum toxin acompared with stretching casts in the treatment of spastic equinus: A randomised prospective trial,” J.Pediatr. Orthop., vol. 18, no. 3, pp. 304–311, 1998. PMID: 9600553

32. Corry I. S., Cosgrove A. P., Duffy C. M., Taylor T. C., and GrahamH. K., “Botulinum toxin A in hamstringspasticity,”Gait Posture, vol. 10, no. 3, pp. 206–210, 1999. PMID: 10567752

33. Desloovere K., Molenaers G., De Cat J., Pauwels P., Van Campenhout A., Ortibus E., et al., “Motorfunction following multilevel botulinum toxin type A treatment in children with cerebral palsy.,” Dev.Med. Child Neurol., vol. 49, no. 1, pp. 56–61, Jan. 2007. PMID: 17209978

34. Detrembleur C., Lejeune T. M., Renders A., and Van Den Bergh P. Y. K., “Botulinum toxin and short-term electrical stimulation in the treatment of equinus in cerebral palsy.,”Mov. Disord., vol. 17, no.1, pp. 162–9, Jan. 2002. PMID: 11835456

35. Galli M., Crivellini M., Santambrogio G. C., Fazzi E., and Motta F., “Short-term effects of ‘botulinumtoxin A’ as treatment for children with cerebral palsy: Kinematic and kinetic aspects at the ankle joint,”Funct. Neurol., vol. 16, no. 4, pp. 317–323, 2001. PMID: 11853322

36. Kay R. M., Rethlefsen S. A., Fern-Buneo A., Wren T. A. L., and Skaggs D. L., “Botulinum toxin as anadjunct to serial casting treatment in children with cerebral palsy,” J. Bone Jt. Surg.—Ser. A, vol. 86,no. 11, pp. 2377–2384, 2004.

37. Klotz M. C. M., Wolf S. I., Heitzmann D., Gantz S., Braatz F., and Dreher T., “The influence of botulinumtoxin a injections into the calf muscles on genu recurvatum in children with cerebral palsy pediatrics,”




http://dx.doi.org/10.1007/s00702-008-0175-8



http://dx.doi.org/10.1007/s11832-013-0511-x


http://dx.doi.org/10.1016/j.pmrj.2014.04.006


http://dx.doi.org/10.1080/10255842.2010.527837











Clin. Orthop. Relat. Res., vol. 471, no. 7, pp. 2327–2332, 2013. doi: 10.1007/s11999-013-2897-7PMID: 23463290

38. Lee J. H., Sung I. Y., Yoo J. Y., Park E. H., Park S. R., “Effects of different dilutions of botulinum toxintype A treatment for children with cerebral palsy with spastic ankle plantarflexor: A randomized con-trolled trial,” J. Rehabil. Med., vol. 41, no. 9, pp. 740–745, Sep. 2009. doi: 10.2340/16501977-0418PMID: 19774308

39. Metaxiotis D., Siebel A., and Doederlein L., “Repeated botulinum toxin A injections in the treatment ofspastic equinus foot.,” Clin. Orthop. Relat. Res., no. 394, pp. 177–85, Jan. 2002.

40. Papadonikolakis A. S., Vekris M. D., V Korompilias A., Kostas J. P., Ristanis S. E., and Soucacos P. N.,“Botulinum A toxin for treatment of lower limb spasticity in cerebral palsy: Gait analysis in 49 patients,”Acta Orthop. Scand., vol. 74, no. 6, pp. 749–755, 2003. PMID: 14763710

41. Polak F., Morton R., Ward C., WallaceW. A., Doderlein L., and Siebel A., “Double-blind comparisonstudy of two doses of botulinum toxin A injected into calf muscles in children with hemiplegic cerebralpalsy.,” Dev. Med. Child Neurol., vol. 44, no. 8, pp. 551–5, Aug. 2002. PMID: 12206622

42. Sutherland D. H., Kaufman K. R., Wyatt M. P., and Chambers H. G., “Injection of botulinum A toxin intothe gastrocnemius muscle of patients with cerebral palsy: A 3-dimensional motion analysis study,”GaitPosture, vol. 4, no. 4, pp. 269–279, 1996.

43. Sutherland D. H., Kaufman K. R., Wyatt M. P., Chambers H. G., and Mubarak S. J., “Double-blind studyof botulinum A toxin injections into the gastrocnemius muscle in patients with cerebral palsy,”Gait Pos-ture, vol. 10, no. 1, pp. 1–9, 1999. PMID: 10469936

44. Svehlik M., Kraus T., Steinwender G., Zwick E. B., Ladecky M., Szabo Z., et al., “Repeated MultilevelBotulinum Toxin A Treatment Maintains Long-TermWalking Ability in Children with Cerebral Palsy,”Ces. A Slov. Neurol. A Neurochir., vol. 75, no. 6, pp. 737–741, 2012.

45. Thompson N. S., Baker R. J., Cosgrove A. P., Corry I. S., and Graham H. K., “Musculoskeletal model-ling in determining the effect of botulinum toxin on the hamstrings of patients with crouch gait,” Dev.Med. Child Neurol., vol. 40, no. 9, pp. 622–625, 1998. PMID: 9766740

46. Wissel J., Muller J., Baldauf A., Ung S. C., Ndayisaba J. P., Stockl B., et al., “Gait analysis to assess theeffects of botulinum toxin type A treatment in cerebral palsy: an open-label study in 10 children withequinus gait pattern,” Eur. J. Neurol., vol. 6, no. 4, pp. S63–S67, Nov. 1999.

47. Zurcher A. W., Molenaers G., Desloovere K., and Fabry G., “Kinematic and kinetic evaluation of theankle after intramuscular injection of botulinum toxin A in children with cerebral palsy,” Acta Orthop.Belg., vol. 67, no. 5, pp. 475–480, 2001. PMID: 11822078



http://dx.doi.org/10.1007/s11999-013-2897-7


http://dx.doi.org/10.2340/16501977-0418







researcharticle literaturereviewandcomparisonoftwo ...botox1(allergan, inc. irvine,...

Documents